Real-Time Handwriting Recognition Interface for Teaching and Learning of Languages in a Multilingual Society
Central Queensland University
ABSTRACT
This paper details the technical design and implementation of a real-time multilingual handwriting recognition interface for teaching and learning of languages in a multilingual society. This implementation allows multiple handwritten languages using Romanised characters (e.g. English and Malay) and ideographs (e.g. Chinese and Japanese) to coexist on a single system for teaching of handwriting. The paper describes how characters written in any language can be dissected into segments of basic strokes. This is mathematically proven that each character stroke, in any language used, have a numerical property that can be uniquely encoded. Based on this encoding, a handwriting teaching software utilising this method is capable of determining if a student is writing a character accurately in the specific language being taught. The advantage of this encoding are: (1) It can be implemented independent of most graphemes and most logographies type written languages, (2) The algorithm is technically accessible and implementable using simple web based scripts such as HTML5, CSS3 and JavaScript, (3) Character accuracy is based on written dynamic stroke-accuracy, and not on static character-pattern. The paper will describe the algorithm in detail and a complete example is presented. A description of a functioning high-fidelity prototype is presented.
KEYWORDS
Handwriting Recognition, Handwritten Characters, Multilingual, Graphemes, Logographies.